#!/usr/bin/perl $sent=0; # sentence-id ################################################################################# #Florest conversion to tigerXML format # #Originly written by Eckhard Bick, eckhard.bick@mail.dk, http://beta.visl.sdu.dk # #Modified by Rui Vilela, ruivilela@di.uminho.pt, http://linguateca.di.uminho.pt ################################################################################# sub to_html{ $_=$_[0]; s/^\s+//;s/\s+$//; s/&/&/g;s//>/; return $_; } sub tags_errors{ $_=$_[0]; s/^\s+//;s/\s+$//; s/\\\'/\'/g; s/\\$/\(/g; s/\\$/\)/g; s/[><]//g;s/&/&/g;s/\"/"/g; return $_; } sub calc_extra{ $_=$_[0]; if ( /$[^<>]*<([^<>]+)>(.*)$/ ){$a=$1;$v=$2; if ( $v=~/[^<>]*<([^<>]+)>(.*)/){$b=$1;$v=$2; if ( $v=~/[^<>]*<([^<>]+)>.*/ ){$c=$1; return "extra=\"".$a."\" extra2=\"".$b."\" extra3=\"".$c."\""; }else{return "extra=\"".$a."\" extra2=\"".$b."\" extra3=\"--\"";} }else{return "extra=\"".$a."\" extra2=\"--\" extra3=\"--\"";} }else{return "extra=\"--\" extra2=\"--\" extra3=\"--\"";} } print "<\?xml version=\"1.0\" encoding=\"ISO-8859-1\"\?>\n"; print "\n"; print "\nQWERTYUIOP\n\n"; ###String para separação print "\n"; while (<>) { next if (/^(\s)?<.*>(\s)*$/); s/^&&$/ /; if (/^SOURCE: (.*)$/) { #Fonte $source =$1; $sent++; $sentslut=0; $source =~ s/\"//g; } elsif (/^(C[PF]\d+\-\d+)\s+(.*)$/) { #Identificação $sentref=$1; $text=$2; $text=~s/\"/"/g; $text=~s/>/>/g; $text=~s/1){ $sent++;$sentslut=0; #$source é igual ; ambiguidade global alternativa da árvore } print '' . "\n"; print '' . "\n"; print '' . "\n"; $t=0; $nt=0; $terminals++; #if ($terminals > $terminalsend +1) {$difference =$terminals-$terminalsend; print "\n!!! $difference\n";} } elsif (/^\s*$/ && $t > 0 && $sentslut==0) { #linha em branco, imprime código XML ou ignora a linha $sentslut=1; # don't print twice if more than one empty line, eller "rubbish" in between sentence and next SOURCE &printxml; } else { m/[^=]/g; $depth = pos; if ($t==0 && $nt==0) { # top node $depth =1; #alterar para 0 para suporte de ad.visl $mother=0; }else{ $mother=$spawning{$depth-1}; } if ( (/\)\t([^ \n]+)/) && ($_!~/$<\w+>$/) ){ # Extrair palavra , pode conter etiqueta secundária,(CJT:? ()) $t++; $word=$1; m/^=*([A-Za-z\-<>\?\&\/]+(?:\[[\+-]\d\])?):([^$\t]+)\(\'([^ \t]+)\'[ $]/; #[-+n] e / Ambiguidade local e de função $func=$1; $pos=$2; $lemma=$3; $func=~s/\/[A-Z<>]+//; #Ignorar ambiguidade local $func=~s/\[[\+-]\d\]//; #Ignorar ambiguidade de função if ($func =~ s/^-//) {$disright=1; $disright{$depth}=1;} #Descontinuidade em nós terminais if ($pos =~ s/-$//) {$disleft=1; $left{$func}=$nt+1; $secmother=0;} #descontinuidades #? #if (! $pos || ! $lemma) {print STDERR "$1\t$2\t$3\n";} if (! $pos) {$pos ="xx";} if (! $lemma) {$lemma ="xx";} if (/\'( +<[^>]+>)* +([A-Za-z&][^\)\t]+)\)/) {$morph=$2;} else {$morph ="--";} ############### Multipla informação para o atributo extra $extra=&calc_extra($_); print '' . "\n"; $size{$mother}++; $id ="s$sent\_$t"; $daughter{$mother,$size{$mother}} =$id; $func{$id} =$func; if ($disright{$depth-1} && $secmother) { # secondary edges til far left mother $size{$secmother}++; $daughter{$secmother,$size{$secmother}} ="*$id"; } } elsif ( (/^=*([^A-Z0-9=\n\&]+)\s*$/) && !(/\?\:/) ) { #punctuation, but not ?: $t++ ;$ponto=$1; $size{$mother}++; $id="s$sent\_$t"; $func{$id} ="--"; $daughter{$mother,$size{$mother}} =$id; print '' . "\n"; } elsif (/^(=*)([A-Za-z\-<>\?\&\/]+(?:\[[\+-]\d\])?):([a-zA-Z\-\?\&]+)\s?($.*$)?[ \t]*$/) { #Nós terminais $func =$2; $form =$3; #$more =$4; $func=~s/\/[A-Z<>]+//; #Ignorar ambiguidade local $func=~s/\[[\+-]\d\]//; #Ignorar ambiguidade de função if ($form =~ s/-$//) {$disleft=1; $left{$func}=$nt+1; $secmother=0;} #descontinuidades if ($func =~ s/^-//) {$disright=1; $disright{$depth}=1;} else {$disright=0;} if ($disright && $mother < $left{$func}) { # if there is no higher level par node in between halves, i.e. not in the case of coordination of -P:vp $disright{$depth}=1; $spawning{$depth} =$left{$func}; # evt. secondary mother } else { $nt500 =$mother + 500; $motherid ="s$sent\_$nt500"; if ($disright && $left{$func} && $mother > $left{$func} && ($func{$motherid} =~ /CJT/)) { $secmother =$left{$func}; } else {$disright{$depth}=0;} $nt++; $form{$nt} =$form; #$depth{$nt} =$depth; ?? $spawning{$depth} =$nt; $size{$mother}++; $nt500 =$nt + 500; $id ="s$sent\_$nt500"; $daughter{$mother,$size{$mother}} =$id; $func{$id} =$func; #if ($more) {$more{$id} =$more;print STDERR "$more\n"} #else {$more{$id} = "--";} if ($disright{$depth-1} && $secmother) { # secondary edges til far left mother $size{$secmother}++; $daughter{$secmother,$size{$secmother}} ="*$id"; } } } } } if ($t >0 || $nt>0) {&printxml;} # last sentence possibly without empty line after it print "<\/body>\n"; print "<\/corpus>\n"; sub printxml { print "<\/terminals>\n"; $terminalsend++; print "\n"; print ''; if ($nt) {$id="s$sent\_501";} else {$id="s$sent\_1";} print ''; print ''; for ($i =1; $i <=$nt; $i++) { $nt500 =$i + 500; $id ="s$sent\_$nt500"; print ''; for ($j =1; $j <= $size{$i}; $j++) { $daughter =$daughter{$i,$j} ; if ($daughter =~ s/^\*//) { print ''; } else { print ''; } } print ''; $size{$i} =0; # nonterminal node sizes reset for next sentence } print ''; print ''; print ''."\n\n"; $nt =0; $t =0; $sentslut =1; for ($k =0; $k <= 20; $k++) { $spawning{$k} =0; # reset depth dependent spawning nodes } }